Harmonizing and Integrating Nursing Data into Multidisciplinary Datasets to Evaluate Hospital Care and Readmissions of Older Adults with Alzheimer’s Disease and Alzheimer’s Disease-Related Dementias

项目摘要/摘要 目前，在美国，有580万老年人患有阿尔茨海默病和阿尔茨海默氏症 疾病相关性痴呆(AD/ADRD)。21%患有AD/ADRD的老年人有计划外， 而且通常是可以预防的，在30天内再次住院，估计每个人的费用超过54亿美元 年。临床数据研究网络(CDRN)及其大型复杂数据集的发展，包括 患者和临床水平的数据导致了预测再入院的各种模型的发展。然而， 现有的模型还没有显示出足够的预测能力。整合护理的机会 这些大型数据集的护理计划(即患者问题、目标和干预)可以填补空白并改善 预测模型的准确性。开发支持集成的研究数据基础设施 因此，将护理计划数据转换为CDRN数据集对于理解和改进至关重要 旨在减少AD/ADRD老年人再入院的跨学科护理。在这个项目中，我们 建议扩大One佛罗里达临床研究联盟和数据信托的研究基础设施 通过第一次创建可重复使用和可行的数据管道，最终将关键字 护理计划数据元素由护士记录到One佛罗里达数据信托中。我们的长期目标是 研究计划是为了更深入地了解AD/ADRD老年人口的再入院情况 通过提供包含跨学科数据的扩展数据集。我们计划进行以下工作 目标：具体目标1(R21阶段)：开发和测试用于提取、翻译和 将佛罗里达大学(UF Health)护理计划数据中的问题和目标整合到 全州范围的One佛罗里达数据信托公司。我们将映射用于表示两个护理数据的本地词汇 使用自然语言处理的国家认可的术语集的要素，并已注册 护士对标测结果进行验证。然后我们将使用自动脚本将本地词汇表替换为 标准化条款，并将数据整合到信托中。具体目标2(R33阶段)：转换和 将UF Health的护理干预整合到Trust中。具体目标3(R33阶段)：提取、转换、 并使用R21阶段的管道将来自第二个组织的护理数据集成到Trust。 具体目标4(R33阶段)：为老年人(65岁及以上)构建和测试再入院预测模型 以上)，与AD/ADRD使用来自信任和环境数据集的相关变量与其链接。我们会 开发机器学习模型，用于从护理要素、其他患者- 信托基金的水平数据，以及影响出院后随访护理的重要环境变量。我们的 该团队将是首批开发可持续数据管道以转换当地护理计划数据的团队之一 转换成标准化数据，以便集成到大型数据集中。通过这项研究，我们将了解到改进措施 护士、医生和其他提供者可以防止再次入院。

Project Summary/Abstract Currently, in the United States, there are 5.8 million older adults with Alzheimer’s Disease and Alzheimer’s Disease-Related Dementias (AD/ADRD). Twenty-one percent of older adults with AD/ADRD have unplanned, and often preventable, hospital readmissions within 30-days, which are estimated to cost over $5.4 billion per year. The development of clinical data research networks (CDRN) and their large complex datasets including patient and clinical level data have led to the development of various models to predict readmissions. However, the existing models have not demonstrated adequate predictive capability. The opportunity to integrate nursing care plan (i.e., patient problems, goals, and interventions) to these large datasets can fill gaps and improve the accuracy of prediction models. The development of a research data infrastructure that supports the integration of nursing care plan data to CDRN datasets is therefore critical for understanding and improving interdisciplinary care aimed at reducing readmissions of older adults with AD/ADRD. In this project, we propose to expand the research infrastructure of the OneFlorida Clinical Research Consortium and Data Trust through the creation, for the first time, of a reusable and feasible data pipeline that will ultimately integrate key care plan data elements documented by nurses into the OneFlorida Data Trust. The long-term goal of our research program is to gain a deeper understanding of readmissions for the aging population with AD/ADRD through the availability of an expanded dataset with interdisciplinary data. We plan to carry out the following aims: Specific Aim 1 (R21 Phase): Develop and test a prototype pipeline for extracting, translating, and integrating problems and goals in nursing care plan data from the University of Florida (UF Health) into the statewide OneFlorida Data Trust. We will map the local vocabulary used to represent the two nursing data elements to nationally recognized terminology sets using natural language processing and have registered nurses validate the mapping results. We will then use automated scripts to replace the local vocabulary with the standardized terms and integrate the data into the Trust. Specific Aim 2 (R33 Phase): Convert and integrate into the Trust nursing interventions from UF Health. Specific Aim 3 (R33 Phase): Extract, convert, and integrate to the Trust nursing data from a second organization, using the pipeline from the R21 phase. Specific Aim 4 (R33 Phase): Construct and test readmissions prediction models for older adults (aged 65 and above) with AD/ADRD using pertinent variables from the Trust and environmental dataset linked to it. We will develop machine-learning models for predicting readmissions from the nursing care elements, other patient- level data in the Trust, and important environmental variables that influence post-discharge follow-up care. Our team will be among the first to develop sustainable data pipelines for transforming local nursing care plan data into standardized data for integration into a large dataset. Through this study, we will learn the improvements nurses, physicians, and other providers can make to prevent readmissions.